Gene expression informatics with an automatic histogram-type membership function for non-uniform data

Akito Daiba, Satoru Ito, Tsutomu Takeuchi, Masafumi Yohda

Research output: Contribution to journalArticle


The non-uniformity of gene expression data is one of the factors that make gene expression analysis difficult. Gene expression data often do not follow a normal distribution but rather various distributions within each group. Thus, it is impossible to apply basic statistical techniques such as the t-test. In this study, we have developed an analysis method for gene expression data obtained by microarrays using a fuzzy logic algorithm with original membership functions. The method automatically evaluates the data from a histogram of gene expression information for a patient group. Using this method, we predicted the efficacy of an anti-TNF-α treatment for rheumatoid arthritis. We created a prediction model for the effects of 14 weeks of anti-TNF-α treatment based on the gene expression data from the peripheral blood of rheumatoid arthritis patients before the treatment. The model had a predictive success of 89% in the model-establishing data group, 94% in the training group, and 89% in the validation group. The results suggest that the method presented here could be an extremely effective tool for gene expression analysis.

Original languageEnglish
Pages (from-to)13-23
Number of pages11
JournalChem-Bio Informatics Journal
Issue number1
Publication statusPublished - 2010 Jan 1



  • Fuzzy logic
  • Gene expression
  • Microarray
  • Prediction of therapeutic efficacy
  • Rheumatoid arthritis

ASJC Scopus subject areas

  • Biochemistry

Cite this